1 Landschaftsökologie Fogwater fluxes above a subtropical montane cloud forest Inauguraldissertation zur Erlangung des Doktorgrades der Naturwissenschaften im Fachbereich Geowissenschaften der Mathematisch-Naturwissenschaftlichen Fakultät der Westfälischen Wilhelms-Universität Münster Vorgelegt von Eva Beiderwieden aus Mülheim an der Ruhr 2007
2 Dekan: Erster Gutachter: Zweiter Gutachter: Prof. Dr. Hans Kerp Prof. Dr. Otto Klemm Prof. Dr. Manfred Krieter Tag der mündlichen Prüfung: Tag der Promotion:
5 Table of contents Table of contents...v List of figures...vii List of tables... x Chapter 1: Introduction Motivation for the project Objectives of this study Structure of this thesis Working hypotheses Fog water deposition... 5 Chapter 2: The impact of fog on the energy budget of a subtropical cypress forest in Taiwan Introduction Materials and methods Results Discussion Conclusions Chapter 3: It goes both ways: measurements of simultaneous evapotranspiration and fog droplet deposition at a montane cloud forest Introduction Methods Results and discussion Conclusions Chapter 4: Nutrient input through occult and wet deposition into a subtropical montane cloud forest Introduction Experimental Results Discussion Conclusions... 67
6 vi Chapter 5: Turbulent fogwater fluxes throughout fog events at a subtropical montane cloud forest in Taiwan Introduction Methods Results and discussion Conclusions...76 Chapter 6: The influence of fog on the turbulent vertical fluxes of CO 2 and water vapor at a subtropical montane cloud forest ecosystem in Taiwan Introduction Materials and methods Results and discussion Conclusions...83 Chapter 7: Synthesis and concluding remarks...85 Summary...89 Zusammenfassung...93 References...97 Danksagung Curriculum Vitae...109
7 vii List of figures Figure 1.1: Aspect of the homogeneous cypress plantation at the Chilan research site... 3 Figure 1.2: Plugging diagram of the instrumentation used to study the fogwater deposition at the Chilan research site Figure 1.3: Meteorological tower at the Chilan research station Figure 1.4: Front and back of the fog droplet spectrometer (FM-100) Figure 1.5: Ultrasonic anemometer (Young 81000) to measure the threedimensional wind velocity Figure 1.6: Active fogwater collector mounted at the meteorological tower Figure 2.1: Relative distribution of wind direction [%] at the experimental site Chilan measured from March through November Grey shaded area: average wind direction of the entire period; transparent area: average wind direction during fog events Figure 2.2: Distribution of the diurnal cycle of fog occurrence at the Chilan site from September through November 2005 using 10-minutes averages Figure 2.3: Typical diurnal variations of visibility and wind direction (top panel), energy balance, soil heat flux, radiation balance, and incoming shortwave radiation (second panel), latent heat, sensible heat, and temperature (third panel), and the CO 2 flux (bottom panel) during two clear days (03/11/2005 and 04/11/2005, Chilan site) Figure 2.4: Typical diurnal variations of visibility and wind direction (top panel), energy balance, soil heat flux, radiation balance, and incoming shortwave radiation (second panel), latent heat, sensible heat, and temperature (third panel), and the CO 2 flux (bottom panel) during two foggy days (01/09/2002 and 02/09/2002, Chilan site) Figure 2.5: Ratio of the sum of turbulent fluxes E, H, and S to available energy Rn during clear conditions at daytime (6 a.m. to 6 p.m.), foggy conditions at daytime, clear conditions at night time (6 p.m. to 6 a.m.), and foggy conditions at night time. The 1:1-line and the correlation coefficient r are given in each section Figure 3.1: (A) Individual diurnal water vapor flux measurements [kg m -2 s -1 ] and mean diurnal water vapor flux Q w [kg m -2 s -1 ], and (B) mean diurnal sensible heat flux Q H [W m -2 ], latent heat flux Q E [W m -2 ], and net radiation Rn [W m -2 ], for the experimental period from September through November Figure 3.2: Average energy fluxes (sensible heat flux Q H [W m -2 ], latent heat flux Q E [W m -2 ]), and (B) water vapor flux Q W [g m -2 s -1 ] for the prevailing wind directions... 40
8 viii Figure 3.3: Courses of (A) turbulent fog water flux Q F [g m -2 s -1 ], water vapor flux Q W [g m -2 s -1 ], (B) sensible heat flux Q H [W m -2 ], latent heat flux Q E [W m -2 ], incoming shortwave radiation Q down [W m -2 ], energy balance EB [W m -2 ], (C) air temperature T (15.06 m and m) [ C], wind direction WD [ ], (D) absolute humidity a [g m -3 ], and the liquid water content LWC [g m -3 ] during a representative fog event measured at the Chilan site ( , 6:00 to 17:30 local time). The turbulent fog water flux and the water vapor flux were measured by using the eddy covariance method. Q W and Q E translate into each other by use of the factor 2465 (with the employed units). Vertical lines indicate times of interest as discussed in the text Figure 3.4: Relationships between turbulent fog water flux [mg m -2 s -1 ] and fog droplet diameter [µm] over time for a representative fog event, using a 30 minute averaging interval on data measured with the eddy covariance method at the Chilan site ( , 6:00 to 17:30 local time Figure 4.1: Schematic diagram of the layout of the neural network used for the fog water flux reproduction. The neurons belonging to the first input layer transmit the values to the neurons in the second layer. After summation of all inputs in every neuron, the weighted sum is transformed by a logistic function giving the output of the neuron. Thus, the values building the inputs of the third layer s neurons are transformed non-linearly when finally summed and again transformed linearly or nonlinearly in the output layer (see text)...52 Figure 4.2: Backward trajectories representing the last 120 hours before reaching the Chilan site ( N and E). A: All trajectories categorized as class I (n = 102). B: All trajectories categorized as class II (n = 20). C: All trajectories categorized as class III (n = 95)...57 Figure 4.3: Median droplet size distribution of droplet number n (black line) and liquid water content LWC (grey line) on the basis of the medians of all 30- minutes intervals with foggy conditions (visibility < 1000 m) during 4 August 2006 through 20 September 2006 at the Chilan site Figure 4.4: Pattern of fog water deposition [mm] during 27 August and 28 August 2006 as an instance for gap filling by using a neural network. The black line represents the originally measured fog water deposition and the grey line is the reproduced fog water deposition...61 Figure 4.5: Nutrient input through occult deposition [mg m -2 ] subdivided into 34 single fog events. A negative deposition means nutrient input into the ecosystem and positive deposition means emission of nutrients as a result of positive fog water fluxes Figure 4.6: Nutrient input through wet deposition [mg m -2 ] subdivided into 20 single rain events...62
9 Figure 4.7: Box-Whisker-plots showing the 5 % percentile, 25 % percentile, 50 % percentile (median), 75 % percentile, and the 95 % percentile of the electric conductivity [µs cm -1 ] (A), ph (B), the concentration of NH 4 + [µeq L -1 ] (C), and the concentration of SO 4 2- [µeq L -1 ] (D) for class I, class II, and class III Figure 5.1: Median droplet size distribution of droplet number n (black line) and liquid water content LWC (grey line) on the basis of the medians of all 30-min intervals with foggy conditions (visibility < 1000 m) Figure 5.2: Pattern of a representative fog event representing the turbulent fogwater flux (A), the energy balance (B), the latent heat flux (C), and the radiation balance (D) during a fog event Figure 5.3: Relationship between the turbulent fogwater flux and the droplet diameter over time for an exemplary fog event, using 30-min intervals of data measured by means of the eddy covariance method at the Chilan site (2006/08/18, 12:00 to 20:00 hours local time) Figure 6.1 Map of the study site with the experimental tower (at 1683 m a.s.l.; N, E) Figure 6.2: Averaged day course for CO 2 flux, water vapor flux, and short wave radiation between 4 th August and 20 th September 2006 at the YYL site under foggy (black line) and clear (grey line) conditions Figure 6.3: Averaged day course for wind direction, temperature, and specific humidity between the 4 th August and 20 th September 2006 at the YYL site under foggy (black line / cross) and clear (grey line / circle) conditions ix
10 x List of tables Table 2.1: Mean values of radiation balance Rn, latent heat flux E, sensible heat flux H, soil heat flux S, sum of E+H+S, energy balance residual EB, energy balance closure EBC, and energy balance ratio EBR measured at the Chilan site...25 Table 2.2: Relation of Rn to E, H, and S under clear and foggy conditions at daytimes and at night...25 Table 3.1: Instrumentation used to measure water vapor, sensible heat, and latent heat fluxes over a two month period in a subtropical montane cloud forest in Taiwan...34 Table 3.2: Sensible heat fluxes Q H, latent heat fluxes Q E, and water vapor fluxes Q W determined with the Bowen Ratio method categorized by wind direction and meteorological conditions Table 4.1: Methods and used instruments for the chemical analysis of the fog and rainwater water samples...54 Table 4.2: Statistical parameters of electric conductivity [µs cm -1 ], ph, and measured ions [µeq L -1 ] of all fogwater and rainwater samples collected between 4 August and 20 September 2006 at the Chilan site. b.d.l. means below detection limit, σ is the standard deviation...55 Table 4.3: Median values, averages and standard deviation of the particular classes. Ion concentrations are given in unit µeq L -1 and electric conductivity is given in µs cm Table 4.4: Significance of the differences within the group of all fogwater samples tested using a one-way ANOVA (electric conductivity, ph, NH 4 +, Ca 2+, Mg 2+, Cl -, NO 3 -, PO 4 3-, SO 4 2-, and F - ) and a Kruskal-Wallis ANOVA (K + and Ca 2+ ). The level of significance was termed after: n.s. - not significant, * - significant (p < 0.05), ** - highly significant (p < 0.01), *** - extremely significant (p < 0.001). For PO 4 3- and F -, the ANOVA did not achieve a result due to the limited data set of class II...58 Table 4.5: Differences between the classes (determined on the basis of the trajectories) tested with a Tukey test (electric conductivity, ph, NH 4 +, Ca 2+, Mg 2+, Cl -, NO 3 -, PO 4 3-, SO 4 2-, and F - ) and a Man-Whitney test (K + and Ca 2+ ). The level of significance was termed after: n.s. - not significant, * - significant (p < 0.05), ** - highly significant (p < 0.01), *** - extremely significant (p < 0.001) Table 4.6: Total nutrient input through occult and wet depsition [mg m -2 ] measured from 04 August to 20 September 2006 at the Chilan site ( n.d. means no data)...63
11 Table 5.1: Relative distribution of the steady-state conditions and the integral turbulence characteristic of the vertical wind during the experimental period (classification after Foken ). Classes 1 to 3 exhibit the highest data quality Table 5.2: Relative distribution of the friction velocity during the experimental period Table 6.1: Averaged fluxes, radiation, specific humidity, temperature, and wind direction from 12:00 to 21:00 hrs local time during the experimental period from 4 th August through 20 th September xi
13 Chapter 1 1 Introduction 1.1 Motivation for the project Montane cloud forest ecosystems Montane cloud forests are tropical or subtropical evergreen forests that are frequently covered by cloud or mist (Stadtmüller 1987). They typically occur at an altitudinal range between 1200 m and 2500 m above sea level where cloud belts are formed by moist ascending air masses. The average annual rainfall is greater than 2500 mm and the air humidity is mostly near its saturation point. The daily temperature ranges between 12 C and 21 C depending on latitude, altitude, aspect, and exposure (Zadroga 1981). Montane cloud forests are characterized by an extremely high biodiversity with respect to herbs, shrubs, and epiphytes, including high rates of endemic species. The trees of montane cloud forests are generally shorter and heavily stemmed, and the high air humidity promotes the development of mosses, bryophytes, and ferns. Epiphytes play an ecologically important role in cloud interception processes. In addition to that, epiphytes such as tank bromeliads and moss balls are recognized to have very high water storage capacities and to release the gathered water slowly (Bruijnzeel 2001). The soils of montane cloud forests are typically wet and close to saturation resulting in slow decomposition of organic matter and a peaty and acidic topsoil (Bruijnzeel and Proctor 1995). The hydrological significance of montane cloud forests is, among other factors, due to the net gain of moisture through cloud water interception by the vegetation and the reduction of evapotranspiration (Hutley et al. 1997). As fog diminishes the solar radiation reaching the vegetation, and the saturation deficit of the foggy air is extremely small, the transpiration rates of montane cloud forests are low (Zadroga 1981). During fog events, the temperature gradients in the canopy profile tend to be more uniform, stable, and generally cooler which might also affect the plant s physiology (Hutley 1997). Cloud forests became one of the most rapidly disappearing forest ecosystems as the land is continuously being deforested and converted to for example cropland and grazing land (Zadroga 1981; Bruijnzeel and Hamilton 2000; Bruijnzeel 2001; Cayuela et al. 2006). Furthermore, montane cloud forests are threatened by global warming of the atmosphere (Bruijnzeel 2001). The lifting of cloud condensation level affects the
14 2 Introduction hydrological cycle and has an impact on the ecological scale as well, since the organisms are subjected to extreme changes of their living conditions. The recognition of montane cloud forests as endangered ecosystems of high ecological significance induced the establishment of several networks such as the Tropical Montane Cloud Forest Initiative founded by the World Conservation Union, WWF International, the World Conservation Monitoring Centre, and the UNESCO International Hydrological Programme. The objectives of this initiative are the implementation of an international cooperation on cloud forest conservation and research activities with a special emphasis on the preservation of water catchments and biodiversity (Bruijnzeel 2001) The Chilan research site The measurements for this study were carried out at the Chilan research station at N and E at an altitude of 1650 m above sea level next to the Yuan Yuang Lake nature preserve in north-eastern Taiwan. The area around the study site consists of an old-grown cypress forest (the yellow cypress Chamaecyparis obtusa var. formosana and the red cypress Chamaecyparis formosenis) which constitutes one of the highquality wood production areas of Taiwan. Around the 1960s, a patch of the old growth was cut down, and the stands were left for regeneration. Within that regenerated plantation, a 1 ha stand was established as a research site in 2002 (Chang et al. 2006). The Chilan study site is part of the long-term ecological research program sponsored by the Taiwan National Science Council in cooperation with the Taiwan Forest Research Institute, the Institute of Botany of the Academia Sinica in Taipeh, the National Dong Hwa University, and other universities. The goal of the program is to assess the biogeochemical cycles of the ecosystem with an emphasis on the role of fog deposition and its effects on the endemic cypress forest. The yellow cypress is the predominant tree species. It accounts for 82 % of the total basal area and has a density of 1820 stems per hectare. The average tree height is 13.7 m and the cypress plantation is characterized by a homogeneous canopy structure (Figure 1.1). The study site was chosen as a long-term research site due to the even-aged stand and the relatively homogeneous topography (the plantation slopes gently with an angle of 14 towards southeast), which is rare at that altitude in Taiwan (Chang et al. 2006). Hsia et al. (2004) point out that due to its excellent fetch conditions, the Chilan research site is well suited to study atmospherecanopy-interactions using micrometeorological methods. Fog occurs almost every day at the study site. The highest duration of fog with an average of 14.2 hours per day is observed in November, and the lowest fog duration is found in July with 2.7 hours per day (Chang et al. 2006). Liao et al. (2003a) report that the reduction of solar radiation through fog droplets as well as the constant exposure to the acidic fog water represent important ecological factors for the growth of the cypress forest.
15 Introduction 3 The climate of the study site is characterized as temperate heavy moist (Chou et al. 2000). The air temperature is relatively low with an annual mean of 13 C. The annual rainfall is on average greater than 4000 mm (Hwang et al. 1996). Much effort has been taken to study the distribution of the cypress forest since both Chameacyparis species are highly appreciated in Taiwan (e.g., Su 1984; Jen 1995). On the one hand, the hardwood is of great economic value due to its strong resistance against termites and fungi (Kuo et al. 2004) and its typical purple-pink coloring (red cypress). On the other hand, the wood has symbolic significance in the Taiwanese heritage and culture. According to pollen analysis, the cypress forest around the Yuan Yuang Lake nature preserve exists for at least 4000 years (Chen and Wu 1999). Figure 1.1: Aspect of the homogeneous cypress plantation at the Chilan research site. 1.2 Objectives of this study The scope of this study is to research the ecological conditions of the Yuan Yuang Lake nature preserve with a special emphasis on fog as the key factor determining the vegetation structure. The endemic appearance of both Chamaecyparis species at the mid-altitudes (1000 m to 2000 m above ground level) in the subtropical montane cloud forests of Taiwan is not fully understood. The light environment of the cloud forest is strongly influenced when fog is present due to the reduction of solar radiation. Fog may thus be the decisive factor in the ecological competition of the forest vegetation.
16 4 Introduction In order to investigate the specific role of fog water in the hydrological and nutrient cycle of the montane cloud forest ecosystem, the thesis is focused on the following aspects: Direct measurements of fogwater fluxes from the atmosphere to the forest canopy, the influence of fog on the energy budget of the ecosystem, the collection of fog water samples and its chemical analysis, the quantification of the nutrient input through occult deposition, and the relevance of nutrient input through occult deposition (fog water) compared to wet deposition (rain). 1.3 Structure of this thesis This thesis consists of five papers that all address the question of how fog water influences the ecological processes of a montane cloud forest in Taiwan. The collected papers have been accepted for publication or submitted to peer-reviewed journals or conference proceedings. Each paper emphasizes a different research question. All aspects taken together provide a comprehensive investigation of the impact of fog on the ecological conditions of the cypress forest. Chapter 1 provides a general introduction to the ecosystem montane cloud forest and presents the Yuan Yuang Lake nature preserve where the field work for this study was performed. Furthermore, some fundamentals about the characteristics of atmospheric turbulence are mentioned. Based on that, the eddy covariance method to determine the turbulent fogwater fluxes is explained. The actual introductions to the individual research topics and approaches follow in the respective papers. Chapter 2 deals with the influence of fog water on the energy budget of the Yuan Yuang Lake nature preserve. In the paper, two experimental periods (11 days with clear weather and 5 days with foggy conditions) at the Chilan research site were compared with regard to the energy budget of the ecosystem. Chapter 3 concentrates on the study of simultaneous evapotranspiration and deposition of fog droplets. By using two different methods (the Bowen ratio method and the eddy covariance method), positive water vapor fluxes were measured despite saturated conditions. The interlinking of the water fluxes and the energy budget of the ecosystem is shown by means of a representative fog event. Chapter 4 deals with the nutrient input through occult and wet deposition into the cypress forest. The event based nutrient input during the 47-day experimental period was calculated and evaluated with regard to its relevance for the ecosystem. The path of the air masses during the last 120 hours before reaching the study site was computed by the help of backward trajectories.
17 Introduction 5 Chapter 5 is focused on the pattern of fog events and the quality of the eddy covariance data. The regime of the fog event was analyzed with an emphasis on the energy balance and the radiation balance of the ecosystem. The data quality was evaluated by means of micrometeorological parameters. Chapter 6 concentrates on the influence of fog water on the CO 2 and water vapor fluxes of the forest ecosystem. The response of the CO 2 and water vapor fluxes as a function of the light conditions of the cypress forest was studied. Additionally, the light use efficiency of the Chamaecyparis species and other environmental conditions to perform photosynthesis during foggy conditions were analyzed. Chapter 7 provides a general synthesis of the study. The results of the different research aspects are linked together, and an outlook for further research is given. 1.4 Working hypotheses Considering former studies performed in cloud forests or comparable ecosystems, the following working hypotheses are suggested: The appearance of both endemic Chamaecyparis species is supposedly related to the occurrence of fog. The reduction of solar radiation during fog is assumed to affect the energy budget of the cypress forest. The ion concentrations of fog water are expected to be higher than those of rain. The chemical composition of the fog water is presumably influenced by the pathway of the air masses. The nutrient input through occult deposition is assumed to provide a significant contribution to the nutrient budget of the ecosystem. 1.5 Fog water deposition Fog water Fog water is recognized to be the key factor sustaining the ecological functions of montane cloud forests. It accounts for a significant hydrological and nutrient input to forest ecosystems (Weathers and Likens 1997; Chang et al. 2006) and may influence the biogeochemistry of the ecosystem (Dawson 1998). It has been widely reported that the concentration of chemical compounds is several times higher in fog water than in precipitation (Schemenauer et al. 1995; Igawa et al. 1998; Bridges et al. 2002; Beiderwieden et al. 2005). Many studies have shown that fog water exhibits particularly high concentrations of ions related to anthropogenic activity such as H +, NH 4 +, NO 3 -, and SO 4 2- (e.g., Klemm and Wrzesinsky 2007). The differences of ion concentrations between fog water and rain are assumed to result from the altitude of the formation
18 6 Introduction processes. Fog, also referred to as a stratocumulus cloud with contact to the surface, represents the lower layer of the atmosphere which is more strongly influenced by surface emissions. Rain droplets originate at higher altitudes where the atmosphere is less polluted from ground-based emissions (Bridges et al. 2002). Another reason for the concentration differences may be the droplet size. Rain droplets are normally much larger than fog droplets and therefore they may be more diluted or less concentrated, respectively. The deposition by means of fog droplets has been defined as occult precipitation since it is not recordable by standard rain gauges and is thus not included in precipitation measurements (Dollard et al. 1983). The quantification of fogwater deposition is still a challenge in the field of atmospheric research due to its instrumental requirements and methodological difficulties. Different methodological approaches such as the modeling of fogwater deposition (Lovett 1984; Pahl 1996; Weathers et al. 2000; Chang 2006), the estimation of the fogwater deposition by weighing plants before and after the fog event (Trautner and Eiden 1988; Chang et al. 2002), water balance methods (Bruijnzeel 2001) or other indirect methods were applied to quantify the fogwater deposition. In recent studies, a direct micrometeorological technique to measure the deposition fluxes of fog water has been realized, i.e. the eddy covariance method (Beswick et al 1991; Vong and Kowalski 1995; Kowalski and Vong 1999; Vermeulen et al. 1997) Atmospheric turbulence In the planetary boundary layer, the horizontal transport of quantities such as moisture, heat, momentum, and pollutants is dominated by the mean wind (advection), and in the vertical direction by turbulent exchange. The turbulence of the planetary boundary layer is principally generated by forcings from the ground, such as thermal turbulence due to solar heating of the surface (buoyancy effects as a result of air density variations with height) or mechanical turbulence as a result of frictional drag. The strength of the mechanical turbulence depends on the wind speed and the surface roughness. Turbulent elements are defined as irregular, quasi-random variations or gusts that last for durations of seconds to minutes (Stull 1988). By averaging, for example, the instantaneous horizontal wind speed U measured over a certain time period (e.g., 30 minutes intervals), we can differentiate a mean wind and a turbulent part of the wind: U ( t) U + u' ( t) = (1.1) Turbulent elements are defined as a type of motion (Stull 1988). As temperature, moisture, and atmospheric scalars are carried by these turbulent movements, their measurements can be divided into a mean and a turbulent component: c = c + c' (1.2)
19 Introduction 7 These short term variations (the turbulent component of Equation 1.2) are induced by small-scale swirls of motion, i.e. eddies. The intensity of such eddies depends on the wind velocity, the characteristics of the surface, and the stratification of the atmosphere (Liljequist and Cehak 1990). The superposition of many eddies of various scales accounts for the atmospheric turbulence (Stull 2000) Estimation of fogwater fluxes The deposition of fog water to the ecosystem is based on two processes, the turbulent deposition and the gravitational settling. The importance of both deposition processes depends on the droplet size as well as on the wind speed and the characteristics of the vegetation cover. In general, the turbulent deposition is higher over forest canopies than over grasslands due to the more pronounced roughness of the forest canopy. Additionally, the leaf area index of trees, especially of conifers, is larger than of grass promoting the impaction of fog droplets (Trautner 1988; Thalmann 2001). Since the estimation of fogwater fluxes is the fundamental method used in this study, the basic concept is explained in the following section Gravitational settling Sedimentation, i.e. the gravitational settling of fog droplets, was calculated using the Stoke s sedimentation velocity v s (Beswick 1991) according to v gd ( ρ ρ ) 2 water air s = (1.3) 18η air where g is the acceleration due to gravity [m s -2 ], η the dynamic viscosity [kg m -1 s -1 ], d the droplet diameter [m], and ρ the density [kg m -3 ]. The gravitational settling contribution D sed [kg m -2 s -1 ] was determined by multiplying the sedimentation velocity v s [m s -1 ] by the liquid water content LWC [kg m -3 ]: D = v, LWC (1.4) sed i s i Turbulent deposition The turbulent deposition of fog droplets, driven by the turbulent exchange between the forest canopy and the atmosphere, was calculated by means of the eddy covariance method. The eddy covariance approach is based on the assumption that each air parcel moving due to turbulent motion carries air scalar properties such as momentum, temperature, gases, aerosols, and liquid water, e.g. fog droplets.
20 8 Introduction Applying Reynolds averaging and assuming that w ' = 0 and c ' = 0, the turbulent vertical flux of a scalar c (e.g., the liquid water content) can be expressed as F c = w' c' (1.5) where F c is the turbulent flux of the scalar c [e.g., kg m -2 s -1 ], w is the vertical wind component [m s -1 ], and c is the concentration of the scalar [e.g., kg m -3 ]. The prime denotes the instantaneous turbulent fluctuation of the individual measurement from its average during the 30 minutes interval. The overbar indicates the time average of the 30 minutes interval. The averaging interval has been chosen to cover all relevant turbulence elements. For a valid application of the eddy covariance method, some assumptions need to be considered (Stull 1988; Foken 2003): Steady state condition of the vertical wind component w and the scalar c during the 30 minutes averaging interval (Foken and Wichura 1996), horizontal homogeneity of the underlying surface ensuring that all sensors are located within the same footprint (horizontal exchange processes can thus be neglected and only the vertical transport has to be considered), Taylor hypothesis (turbulence elements pass the sensors as frozen elements, i.e. the horizontal wind speed translates the turbulence elements as a function of time to their corresponding measurement in space) (Stull 1988), and the use of fast-response sensors to quantify the respective scalar (temporal resolution of Hz). If an adequate technique is available, the measurement of the turbulent fluxes by means of the eddy covariance method is possible for all air components. In recent years, appropriate instruments have been developed for air constituents like CO 2 and water vapor (e.g., the LI-COR 7500), ozone, or methane. Using slower sensors, the turbulent components will be filtered by the instrument response, resulting in incorrect fluxes (Stull 1988). Negative fluxes are directed downward and indicate deposition into the ecosystem. In contrast, positive fluxes are directed upward and indicate emission out of the ecosystem Total fogwater deposition The total fogwater deposition D total was calculated by adding the gravitational and the turbulent deposition contributions: D D + D total = (1.6) sed turb The total fogwater input of each 30 minutes interval [kg m -2 ] was derived by multiplying the fog deposition rates [kg m -2 s -1 ] by 1800 s. The nutrient input through fog deposition [mg m -2 ] was estimated by multiplying the total fogwater deposition